AI-powered tools can quickly process enormous amounts of data. They can mine relations and find patterns that humans might miss. However, they are black boxes, which makes it challenging to apply them in a trustworthy manner and to ensure accurate reporting with high source transparency. Our vision as Human-Computer Interaction and AI scholars is to empower journalists and to give them more control over automation. In this interactive workshop, we will present practical use cases where AI and automation could support investigative reporters and underline the challenges journalists must keep an eye on. We want to equip reporters with more structure to automate certain aspects of their investigations. For this, we combine Programming-by-Demonstration (PbD) with the power of large language models. PbD enables end users to automate tasks without programming experience and makes the underlying automation processes transparent. We report our findings on potential applications in journalists' workflows and the practical utility of such automation tools based on in-depth interviews with investigative and data journalists. We further explain how to adapt these automation systems to the way journalists work.

In this interactive workshop, we invite participants to collaborate with us to conceptualize, design, and shape the features of an automated system created by journalists for journalists.